Post subject: De-embedding lossy transmission line
Posted: Wed Jul 25, 2007 4:40 pm
Wed Jul 25, 2007 4:23 pm
Hi all experts ,
I'm doing the following experiment:
I have a system
which consists of a pair of SMA cables, backplane with microstrip
differential pair and another pair of cables which S-parameters
I'm trying to find.
Unfortunately, I can't just measure
S-parameters of the second pair of cables by using a VNA. But I
have S-parameters of the whole system: [SMA cables - backplane -
cables], and S-parameters of [SMA cables] and [backplane] separately.
The backplane is relatively ‘lossy’ at high frequencies (>10GHz).
I'm converting S-parameters to T-parameters, then taking
T[backplane]^-1*T[SMA-cables]^-1*T[whole system]. Then, I’m converting
back to S expecting to get "second" cables S matrix. Beyond some
frequency (~12GHz), I'm getting very high gain and obviously wrong
behavior. I used this technique in the past to de-embed NOT very
What could be wrong?
Thanks a lot
Posted: Thu Jul 26, 2007 3:39 am
Fri Feb 17, 2006 12:07 pm
Location: London UK
One time I encountered something similar, where
a VNA plot turned to rat doo-dooz above a certain frequency, it
turned out to be an unexpected resonance. This produced a rapid
variation in S parameters around the resonant frequency, and at
harmonics of it.
Is there any chance your set-up could exhibit